Burn severity is related to fire intensity and fire duration and provides a quantitative measure related to fire impact and biomass consumption. Traditional field-based methods to estimate burn severity are time consuming, labor intensive, and normally limited in spatial extent. Remotely sensed data may provide a means to estimate severity levels across large areas, but it is critical to understand the causes of variability in spectral response with variations in burn severity. To address this issue, a combined leaf (Prospect) and canopy (Kuusk) reflectance model was used to simulate the spectral response of a range of vegetation canopies with different burn severity levels. The key aspects examined in the simulations were change in soil color, change in foliage color from green to brown (burned), and change in leaf area index (LAI). For each simulation the composite burn index (CBI) was determined using the same rules used in the field to estimate burn severity levels. Statistical analyses examined the strength of the correlations between CBI and reflectance in individual wave bands in the 400–2500 nm range and CBI and a range of spectral indices combining pairs of wave bands. The results showed that wave bands in the near infrared (NIR) were most strongly related to the CBI of the simulated canopies because of their sensitivity to reduction in LAI. Spectral indices combining reflectance in wave bands in the NIR and shortwave infrared and red edge region showed stronger correlations with CBI. Forward stepwise regression with two to six terms selected wave bands in these regions and accounted for more than 90% of the variation in CBI.